eal: add device event monitor framework

This patch aims to add a general device event monitor framework at
EAL device layer, for device hotplug awareness and actions adopted
accordingly. It could also expand for all other types of device event
monitor, but not in this scope at the stage.

To get started, users firstly call below new added APIs to enable/disable
the device event monitor mechanism:
  - rte_dev_event_monitor_start
  - rte_dev_event_monitor_stop

Then users shell register or unregister callbacks through the new added
APIs. Callbacks can be some device specific, or for all devices.
  -rte_dev_event_callback_register
  -rte_dev_event_callback_unregister

Use hotplug case for example, when device hotplug insertion or hotplug
removal, we will get notified from kernel, then call user's callbacks
accordingly to handle it, such as detach or attach the device from the
bus, and could benefit further fail-safe or live-migration.

Signed-off-by: Jeff Guo <jia.guo@intel.com>
Reviewed-by: Jianfeng Tan <jianfeng.tan@intel.com>

doc/guides/rel_notes/release_18_05.rst

@@ -58,6 +58,16 @@ New Features
   * Added support for NVGRE, VXLAN and GENEVE filters in flow API.
   * Added support for DROP action in flow API.
 
+* **Added device event monitor framework.**
+
+  Added a general device event monitor framework at EAL, for device dynamic management.
+  Such as device hotplug awareness and actions adopted accordingly. The list of new APIs:
+
+  * ``rte_dev_event_monitor_start`` and ``rte_dev_event_monitor_stop`` are for
+    the event monitor enable and disable.
+  * ``rte_dev_event_callback_register`` and ``rte_dev_event_callback_unregister``
+    are for the user's callbacks register and unregister.
+
 
 API Changes
 -----------

lib/librte_eal/bsdapp/eal/Makefile

@@ -34,6 +34,7 @@ SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_lcore.c
 SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_timer.c
 SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_interrupts.c
 SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_alarm.c
+SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_dev.c
 
 # from common dir
 SRCS-$(CONFIG_RTE_EXEC_ENV_BSDAPP) += eal_common_lcore.c

lib/librte_eal/bsdapp/eal/eal_dev.c

@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2018 Intel Corporation
+ */
+
+#include <rte_log.h>
+#include <rte_compat.h>
+#include <rte_dev.h>
+
+int __rte_experimental
+rte_dev_event_monitor_start(void)
+{
+	RTE_LOG(ERR, EAL, "Device event is not supported for FreeBSD\n");
+	return -1;
+}
+
+int __rte_experimental
+rte_dev_event_monitor_stop(void)
+{
+	RTE_LOG(ERR, EAL, "Device event is not supported for FreeBSD\n");
+	return -1;
+}

lib/librte_eal/bsdapp/eal/meson.build

@@ -13,4 +13,5 @@ env_sources = files('eal_alarm.c',
 		'eal_timer.c',
 		'eal.c',
 		'eal_memory.c',
+		'eal_dev.c'
 )

lib/librte_eal/common/eal_common_dev.c

@@ -14,9 +14,34 @@
 #include <rte_devargs.h>
 #include <rte_debug.h>
 #include <rte_log.h>
+#include <rte_spinlock.h>
+#include <rte_malloc.h>
 
 #include "eal_private.h"
 
+/**
+ * The device event callback description.
+ *
+ * It contains callback address to be registered by user application,
+ * the pointer to the parameters for callback, and the device name.
+ */
+struct dev_event_callback {
+	TAILQ_ENTRY(dev_event_callback) next; /**< Callbacks list */
+	rte_dev_event_cb_fn cb_fn;            /**< Callback address */
+	void *cb_arg;                         /**< Callback parameter */
+	char *dev_name;	 /**< Callback device name, NULL is for all device */
+	uint32_t active;                      /**< Callback is executing */
+};
+
+/** @internal Structure to keep track of registered callbacks */
+TAILQ_HEAD(dev_event_cb_list, dev_event_callback);
+
+/* The device event callback list for all registered callbacks. */
+static struct dev_event_cb_list dev_event_cbs;
+
+/* spinlock for device callbacks */
+static rte_spinlock_t dev_event_lock = RTE_SPINLOCK_INITIALIZER;
+
 static int cmp_detached_dev_name(const struct rte_device *dev,
 	const void *_name)
 {
@@ -207,3 +232,139 @@ rte_eal_hotplug_remove(const char *busname, const char *devname)
 	rte_eal_devargs_remove(busname, devname);
 	return ret;
 }
+
+int __rte_experimental
+rte_dev_event_callback_register(const char *device_name,
+				rte_dev_event_cb_fn cb_fn,
+				void *cb_arg)
+{
+	struct dev_event_callback *event_cb;
+	int ret;
+
+	if (!cb_fn)
+		return -EINVAL;
+
+	rte_spinlock_lock(&dev_event_lock);
+
+	if (TAILQ_EMPTY(&dev_event_cbs))
+		TAILQ_INIT(&dev_event_cbs);
+
+	TAILQ_FOREACH(event_cb, &dev_event_cbs, next) {
+		if (event_cb->cb_fn == cb_fn && event_cb->cb_arg == cb_arg) {
+			if (device_name == NULL && event_cb->dev_name == NULL)
+				break;
+			if (device_name == NULL || event_cb->dev_name == NULL)
+				continue;
+			if (!strcmp(event_cb->dev_name, device_name))
+				break;
+		}
+	}
+
+	/* create a new callback. */
+	if (event_cb == NULL) {
+		event_cb = malloc(sizeof(struct dev_event_callback));
+		if (event_cb != NULL) {
+			event_cb->cb_fn = cb_fn;
+			event_cb->cb_arg = cb_arg;
+			event_cb->active = 0;
+			if (!device_name) {
+				event_cb->dev_name = NULL;
+			} else {
+				event_cb->dev_name = strdup(device_name);
+				if (event_cb->dev_name == NULL) {
+					ret = -ENOMEM;
+					goto error;
+				}
+			}
+			TAILQ_INSERT_TAIL(&dev_event_cbs, event_cb, next);
+		} else {
+			RTE_LOG(ERR, EAL,
+				"Failed to allocate memory for device "
+				"event callback.");
+			ret = -ENOMEM;
+			goto error;
+		}
+	} else {
+		RTE_LOG(ERR, EAL,
+			"The callback is already exist, no need "
+			"to register again.\n");
+		ret = -EEXIST;
+	}
+
+	rte_spinlock_unlock(&dev_event_lock);
+	return 0;
+error:
+	free(event_cb);
+	rte_spinlock_unlock(&dev_event_lock);
+	return ret;
+}
+
+int __rte_experimental
+rte_dev_event_callback_unregister(const char *device_name,
+				  rte_dev_event_cb_fn cb_fn,
+				  void *cb_arg)
+{
+	int ret = 0;
+	struct dev_event_callback *event_cb, *next;
+
+	if (!cb_fn)
+		return -EINVAL;
+
+	rte_spinlock_lock(&dev_event_lock);
+	/*walk through the callbacks and remove all that match. */
+	for (event_cb = TAILQ_FIRST(&dev_event_cbs); event_cb != NULL;
+	     event_cb = next) {
+
+		next = TAILQ_NEXT(event_cb, next);
+
+		if (device_name != NULL && event_cb->dev_name != NULL) {
+			if (!strcmp(event_cb->dev_name, device_name)) {
+				if (event_cb->cb_fn != cb_fn ||
+				    (cb_arg != (void *)-1 &&
+				    event_cb->cb_arg != cb_arg))
+					continue;
+			}
+		} else if (device_name != NULL) {
+			continue;
+		}
+
+		/*
+		 * if this callback is not executing right now,
+		 * then remove it.
+		 */
+		if (event_cb->active == 0) {
+			TAILQ_REMOVE(&dev_event_cbs, event_cb, next);
+			free(event_cb);
+			ret++;
+		} else {
+			continue;
+		}
+	}
+	rte_spinlock_unlock(&dev_event_lock);
+	return ret;
+}
+
+void
+dev_callback_process(char *device_name, enum rte_dev_event_type event)
+{
+	struct dev_event_callback *cb_lst;
+
+	if (device_name == NULL)
+		return;
+
+	rte_spinlock_lock(&dev_event_lock);
+
+	TAILQ_FOREACH(cb_lst, &dev_event_cbs, next) {
+		if (cb_lst->dev_name) {
+			if (strcmp(cb_lst->dev_name, device_name))
+				continue;
+		}
+		cb_lst->active = 1;
+		rte_spinlock_unlock(&dev_event_lock);
+		cb_lst->cb_fn(device_name, event,
+				cb_lst->cb_arg);
+		rte_spinlock_lock(&dev_event_lock);
+		cb_lst->active = 0;
+	}
+	rte_spinlock_unlock(&dev_event_lock);
+}

lib/librte_eal/common/eal_private.h

@@ -9,6 +9,8 @@
 #include <stdint.h>
 #include <stdio.h>
 
+#include <rte_dev.h>
+
 /**
  * Initialize the memzone subsystem (private to eal).
  *
@@ -238,4 +240,16 @@ struct rte_bus *rte_bus_find_by_device_name(const char *str);
 
 int rte_mp_channel_init(void);
 
+/**
+ * Internal Executes all the user application registered callbacks for
+ * the specific device. It is for DPDK internal user only. User
+ * application should not call it directly.
+ *
+ * @param device_name
+ *  The device name.
+ * @param event
+ *  the device event type.
+ */
+void dev_callback_process(char *device_name, enum rte_dev_event_type event);
+
 #endif /* _EAL_PRIVATE_H_ */

lib/librte_eal/common/include/rte_dev.h

@@ -24,6 +24,25 @@ extern "C" {
 #include <rte_compat.h>
 #include <rte_log.h>
 
+/**
+ * The device event type.
+ */
+enum rte_dev_event_type {
+	RTE_DEV_EVENT_ADD,	/**< device being added */
+	RTE_DEV_EVENT_REMOVE,	/**< device being removed */
+	RTE_DEV_EVENT_MAX	/**< max value of this enum */
+};
+
+struct rte_dev_event {
+	enum rte_dev_event_type type;	/**< device event type */
+	int subsystem;			/**< subsystem id */
+	char *devname;			/**< device name */
+};
+
+typedef void (*rte_dev_event_cb_fn)(char *device_name,
+					enum rte_dev_event_type event,
+					void *cb_arg);
+
 __attribute__((format(printf, 2, 0)))
 static inline void
 rte_pmd_debug_trace(const char *func_name, const char *fmt, ...)
@@ -267,4 +286,78 @@ __attribute__((used)) = str
 }
 #endif
 
+/**
+ * @warning
+ * @b EXPERIMENTAL: this API may change without prior notice
+ *
+ * It registers the callback for the specific device.
+ * Multiple callbacks cal be registered at the same time.
+ *
+ * @param device_name
+ *  The device name, that is the param name of the struct rte_device,
+ *  null value means for all devices.
+ * @param cb_fn
+ *  callback address.
+ * @param cb_arg
+ *  address of parameter for callback.
+ *
+ * @return
+ *  - On success, zero.
+ *  - On failure, a negative value.
+ */
+int __rte_experimental
+rte_dev_event_callback_register(const char *device_name,
+				rte_dev_event_cb_fn cb_fn,
+				void *cb_arg);
+
+/**
+ * @warning
+ * @b EXPERIMENTAL: this API may change without prior notice
+ *
+ * It unregisters the callback according to the specified device.
+ *
+ * @param device_name
+ *  The device name, that is the param name of the struct rte_device,
+ *  null value means for all devices and their callbacks.
+ * @param cb_fn
+ *  callback address.
+ * @param cb_arg
+ *  address of parameter for callback, (void *)-1 means to remove all
+ *  registered which has the same callback address.
+ *
+ * @return
+ *  - On success, return the number of callback entities removed.
+ *  - On failure, a negative value.
+ */
+int __rte_experimental
+rte_dev_event_callback_unregister(const char *device_name,
+				  rte_dev_event_cb_fn cb_fn,
+				  void *cb_arg);
+
+/**
+ * @warning
+ * @b EXPERIMENTAL: this API may change without prior notice
+ *
+ * Start the device event monitoring.
+ *
+ * @return
+ *   - On success, zero.
+ *   - On failure, a negative value.
+ */
+int __rte_experimental
+rte_dev_event_monitor_start(void);
+
+/**
+ * @warning
+ * @b EXPERIMENTAL: this API may change without prior notice
+ *
+ * Stop the device event monitoring.
+ *
+ * @return
+ *   - On success, zero.
+ *   - On failure, a negative value.
+ */
+int __rte_experimental
+rte_dev_event_monitor_stop(void);
+
 #endif /* _RTE_DEV_H_ */

lib/librte_eal/linuxapp/eal/Makefile

@@ -42,6 +42,7 @@ SRCS-$(CONFIG_RTE_EXEC_ENV_LINUXAPP) += eal_lcore.c
 SRCS-$(CONFIG_RTE_EXEC_ENV_LINUXAPP) += eal_timer.c
 SRCS-$(CONFIG_RTE_EXEC_ENV_LINUXAPP) += eal_interrupts.c
 SRCS-$(CONFIG_RTE_EXEC_ENV_LINUXAPP) += eal_alarm.c
+SRCS-$(CONFIG_RTE_EXEC_ENV_LINUXAPP) += eal_dev.c
 
 # from common dir
 SRCS-$(CONFIG_RTE_EXEC_ENV_LINUXAPP) += eal_common_lcore.c

lib/librte_eal/linuxapp/eal/eal_dev.c

@@ -0,0 +1,22 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2018 Intel Corporation
+ */
+
+#include <rte_log.h>
+#include <rte_compat.h>
+#include <rte_dev.h>
+
+
+int __rte_experimental
+rte_dev_event_monitor_start(void)
+{
+	/* TODO: start uevent monitor for linux */
+	return 0;
+}
+
+int __rte_experimental
+rte_dev_event_monitor_stop(void)
+{
+	/* TODO: stop uevent monitor for linux */
+	return 0;
+}

lib/librte_eal/linuxapp/eal/meson.build

@@ -19,6 +19,7 @@ env_sources = files('eal_alarm.c',
 		'eal_vfio_mp_sync.c',
 		'eal.c',
 		'eal_memory.c',
+		'eal_dev.c',
 )
 
 if has_libnuma == 1

lib/librte_eal/rte_eal_version.map

@@ -213,6 +213,10 @@ DPDK_18.02 {
 EXPERIMENTAL {
 	global:
 
+	rte_dev_event_callback_register;
+	rte_dev_event_callback_unregister;
+	rte_dev_event_monitor_start;
+	rte_dev_event_monitor_stop;
 	rte_eal_cleanup;
 	rte_eal_devargs_insert;
 	rte_eal_devargs_parse;